Commercial vehicle longitudinal and transverse cooperative anti-yaw stability control system and method

Abstract

The invention discloses a commercial vehicle longitudinal and transverse cooperative anti-yaw stability control system and method, and aims to solve the problem that a traditional vehicle stability control system is lagged in response and cannot actively prevent the risk of instability. The method comprises the following steps: firstly, constructing a self-adaptive digital twinborn model which can be synchronized with a physical vehicle in real time through an online parameter identification technology; then, forward simulation is carried out based on the high-fidelity model, and the future state trajectory of the vehicle is converted into a quantified stability risk potential field containing multi-dimensional information such as yawing, rollover and hinge folding; the core of the method is that by analyzing the gradient of the risk potential field in the state space, the instability risk mode playing a leading role at present is accurately and dynamically identified. According to the method, fundamental transformation from passive response to active prevention is achieved, and the driving safety and stability of the commercial vehicle under the limiting working condition are remarkably improved through accurate prediction and risk classification.

Description

Technical Field

[0001] This invention relates to the field of vehicle control technology, specifically to a longitudinal and lateral coordinated anti-yaw stability control system and method for commercial vehicles. Background Technology

[0002] With the rapid development of the automotive industry, vehicle driving safety has become one of the core indicators for measuring technological level. For commercial vehicles, especially tractor-semi-trailer combinations with characteristics such as high center of gravity, large load capacity and multiple articulation points, they are prone to instability phenomena such as sideslip, fishtailing, rollover or articulation folding under extreme conditions such as high speed, emergency steering or poor road adhesion, which pose a serious threat to traffic safety.

[0003] To improve vehicle handling stability, active safety technologies such as Electronic Stability Control (ESC) have emerged and become widely used. These traditional stability control systems work by monitoring the driver's intentions (such as steering wheel angle) and the vehicle's actual motion (such as yaw rate and lateral acceleration) in real time. When a significant deviation is detected—that is, when the vehicle's trajectory deviates from the driver's expectations—a corrective yaw moment is generated by applying differential braking force to one or more wheels and coordinating engine torque, thereby helping the vehicle regain stability. Undoubtedly, this type of technology plays a crucial role in preventing vehicle loss of control.

[0004] However, existing technologies generally suffer from an inherent limitation in their underlying logic: they are essentially "passive-response" control. The activation of these systems relies on the real-time capture of instability trends that have already occurred or are occurring. In other words, the control system only intervenes when unstable dynamic characteristics (e.g., excessive sideslip angle or yaw rate error) have accumulated to a level clearly identifiable by the sensor network. In rapidly changing emergency situations, this unavoidable response lag may cause the system to miss the optimal intervention opportunity, significantly reducing the control effect, and in some extreme cases, even becoming irreversible.

[0005] Furthermore, traditional stability control systems fall short in diagnosing the root causes of risks. They typically attribute all instability tendencies to uncontrolled yaw motion and primarily intervene using the relatively singular method of differential braking. This "one-size-fits-all" control strategy struggles to accurately distinguish whether an emerging risk stems from simple tire sideslip, an impending rollover, or the folding tendency inherent in articulated trains. To address risks with different root causes, the optimal control strategy should differ: suppressing sideslip requires precise yaw moments, avoiding rollover necessitates rapid reduction of speed and lateral acceleration, and preventing folding may require diametrically opposed longitudinal force controls on the tractor and trailer. Current technologies lack the ability to deeply identify risk patterns, therefore their control measures are often not globally optimal solutions.

[0006] Furthermore, commercial vehicles possess highly time-varying physical characteristics. Key parameters such as total mass, center of gravity, and tire wear can change drastically under different loading conditions and usage cycles. Traditional control systems are mostly based on a set of vehicle model parameters fixed in the controller, representing calibrated operating conditions. When the actual parameters of the vehicle deviate significantly from these calibrated parameters, the controller's understanding of the vehicle's state will be flawed, affecting the accuracy of its judgments and the precision of its control, thus severely limiting its adaptability.

[0007] Therefore, how to break through the existing control technology's "passive response" framework and achieve "active prediction" of instability risks; how to accurately identify the fundamental risk patterns from complex dynamic phenomena; and how to adaptively adopt "targeted" collaborative control strategies based on the vehicle's real-time characteristics have become key technical challenges that urgently need to be addressed to improve the active safety performance of heavy commercial vehicles. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to provide a longitudinal and lateral coordinated anti-yaw stability control system and method for commercial vehicles that can overcome the shortcomings of existing commercial vehicle stability control technologies, such as slow response, single control methods at the expense of vehicle kinetic energy, and poor adaptability to time-varying parameters such as vehicle load and road conditions.

[0009] To address the aforementioned technical problems, the first aspect of this invention provides a longitudinal and lateral coordinated anti-yaw stability control system for commercial vehicles. This system breaks through the traditional control approach based on error feedback and establishes an active, adaptive, and coordinated control framework based on risk prediction and mode diagnosis.

[0010] The system includes:

[0011] The adaptive vehicle state digital twin construction module aims to establish an online model that operates synchronously with the physical vehicle and is capable of self-correction. Specifically, this module continuously collects various sensor data from the vehicle and compares it with the output of the built-in vehicle dynamics model. When discrepancies exist, the module uses online estimation algorithms, such as Extended Kalman Filter (EKF), to estimate and update key time-varying parameters characterizing the vehicle's dynamic properties in real time. These key time-varying parameters preferably include vehicle mass, center of gravity position, and tire-road peak adhesion coefficient, ensuring that the model maintains high fidelity despite changes in vehicle load and road conditions, laying the foundation for subsequent accurate predictions.

[0012] The stability risk potential field prediction module is connected to the adaptive vehicle state digital twin construction module. Its core function is to shift from "post-event correction" to "pre-event prevention." This module utilizes the real-time state and parameters provided by the digital twin model, combined with the driver's operational intentions, to predict the vehicle's state trajectory in the future prediction time domain through forward simulation. It does not calculate the current state error but rather the "risk potential energy" for the future state to enter the unstable region. This risk potential energy is constructed as a multi-dimensional stability risk potential field, which is composed of multiple weighted sub-potential fields related to different instability modes. Its preferred calculation method satisfies the following relationship:

[0013] U risk =w yaw U yaw +w roll U roll +w jack U jack ;

[0014] Among them, U risk To comprehensively assess the risk potential, U yaw U represents the yaw risk sub-potential field determined by the yaw angular velocity and the sideslip angle of the center of mass. roll U represents the sub-potential field representing the rollover risk determined by the lateral load transfer rate. jack w represents the articulated folding risk subpotential field determined by the hinge angle. yaw w roll w jack These are the corresponding weighting coefficients.

[0015] The risk mode dynamic identification module, connected to the stability risk potential field prediction module, is responsible for "diagnosing" the risks faced by the system and determining the fundamental type of instability. Specifically, this module analyzes the topological structure of the multidimensional stability risk potential field—that is, calculates its gradient in the vehicle state space—and identifies the dominant instability risk mode based on the directions of the main components of the gradient. This approach allows the system to not only understand the magnitude of the risk but also its nature. The dominant instability risk modes include at least one of the following: yaw instability mode, rollover risk mode, and articulated folding mode.

[0016] The multi-actuator collaborative control module, connected to the risk mode dynamic identification module, is the execution center for achieving targeted and precise control in this system. Based on the identified dominant instability risk modes, it adaptively generates and executes collaborative control strategies. Its adaptability is reflected in its dynamic adjustment of the control target and the priority of calling multiple actuators on the vehicle (such as the drive system, braking system, and auxiliary braking system) for different identified dominant instability risk modes.

[0017] Preferably, when the dominant instability risk mode is the yaw instability mode, the system determines that the most effective intervention is to generate a direct yaw correction torque. Therefore, the cooperative control strategy prioritizes calling the vehicle's drive system (through the torque differential between the left and right wheels) or the differential braking system.

[0018] Preferably, when the dominant instability risk mode is the rollover risk mode, the system determines that the primary task at this time is to reduce the lateral force. Therefore, the cooperative control strategy prioritizes calling the vehicle's auxiliary braking or service braking system to actively reduce the vehicle speed, and actively limits the differential braking intervention that generates asymmetric braking force in order to avoid aggravating the rollover.

[0019] Preferably, when the dominant instability risk mode is the articulated folding mode, the system determines that the articulation angle θ needs to be suppressed. h The divergence of the control strategy leads to a non-obvious stretching strategy, which prioritizes braking the semi-trailer wheels while maintaining or increasing the driving force of the tractor.

[0020] A second aspect of this invention provides a method for longitudinal and lateral coordinated anti-yaw stability control of commercial vehicles, which is achieved through the following steps:

[0021] Construct an adaptive digital twin of vehicle state to estimate key time-varying parameters characterizing vehicle dynamics in real time;

[0022] Based on the key time-varying parameters, the multidimensional stability risk potential field of the vehicle in the future prediction time domain is predicted.

[0023] By analyzing the topological structure of the multidimensional stability risk potential field, the dominant instability risk mode is identified.

[0024] Based on the dominant instability risk pattern, a collaborative control strategy is adaptively generated and executed.

[0025] This invention provides a longitudinal and lateral coordinated anti-yaw stability control system and method for commercial vehicles. It has the following beneficial effects:

[0026] 1. This invention represents a leap from passive response to proactive prevention, significantly enhancing safety margins. By constructing an adaptive digital twin model and performing forward simulation, this invention generates a quantified "stability risk potential field," enabling prediction of instability trends hundreds of milliseconds or even earlier. This predictive capability allows the control system to intervene earlier, taking gentler and more timely intervention measures. This avoids the lag of traditional ESC systems, which only apply emergency braking after a problem occurs, greatly expanding the vehicle's safety boundaries.

[0027] 2. It enables precise diagnosis of the root causes of risks, resulting in more targeted control strategies. This invention uniquely identifies the dominant instability risk mode—whether it's yaw, rollover, or articulated folding—by analyzing the gradient of the risk potential field in the state space. This targeted diagnostic capability allows the system to move beyond the limitations of traditional "one-size-fits-all" control, providing precise decision-making basis for adopting the most efficient control strategy and avoiding ineffective or even negatively impactful control actions.

[0028] 3. Significantly improves the adaptability and robustness of the control system. Utilizing a parameter co-estimation algorithm based on extended Kalman filtering, the digital twin model of this invention can identify time-varying parameters such as the vehicle's total mass, center of gravity position, and tire lateral stiffness online in real time. This means that regardless of whether the vehicle is unloaded, fully loaded, or the tires are worn, the control system can always make predictions and decisions based on a model that most closely approximates the vehicle's actual state, ensuring high consistency and reliability of control performance under various operating conditions.

[0029] 4. By optimizing the coordination of multiple actuators, control efficiency is maximized. This invention adaptively matches different control objectives and actuator call priorities based on the identified different risk patterns. For example, differential braking is prioritized when dealing with yaw risk, and symmetrical deceleration is prioritized when dealing with rollover risk. This intelligent coordination mechanism ensures that at any given time, multiple actuators of the vehicle (such as braking, driving, and steering) can combine their forces in an optimal way, thereby mitigating risks with minimal cost and maximum speed.

[0030] 5. Improved driving experience and control smoothness under extreme conditions. Because this invention can intervene at an early stage of risk occurrence, and its intervention measures are precisely tailored to the root cause of the risk, its control effect is smoother and gentler compared to traditional systems. This not only effectively avoids the risk of severe impact and secondary instability caused by excessive or improper emergency braking, but also allows the driver to experience stronger vehicle controllability and stability under extreme conditions, improving driving confidence and ride comfort. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the functional modules of the control system of the present invention;

[0032] Figure 2 This is an overall flowchart of the control method of the present invention;

[0033] Figure 3 This is the matching logic diagram for the adaptive cooperative control strategy of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0035] refer to Figures 1 to 3 This invention provides a longitudinal and lateral coordinated anti-yaw stability control system and method for commercial vehicles. The system can be integrated into the vehicle's domain controller or a dedicated electronic control unit (ECU). By interacting with the vehicle's sensor network and actuator network, it achieves a high level of active control over vehicle stability.

[0036] In a preferred embodiment, the core logic function of the longitudinal and lateral coordinated anti-yaw stability control system for commercial vehicles can be divided into four tightly coupled modules: an adaptive vehicle state digital twin construction module, a stability risk potential field prediction module, a risk mode dynamic identification module, and a multi-actuator coordinated control module.

[0037] The system's workflow forms a continuously operating intelligent closed loop, from data perception to control execution.

[0038] First, the system's operation begins with comprehensive data acquisition. An onboard sensor network, including standard and extended sensors, continuously inputs real-time dynamic information about the vehicle into the controller. This information includes, but is not limited to: wheel speeds of each wheel, driver's steering wheel angle, yaw rate sensor readings and longitudinal / lateral acceleration sensor readings, brake master cylinder pressure, and more granular data such as air suspension airbag pressure, articulated train articulation angle, electric motor torque and speed of new energy commercial vehicles, and high-precision vehicle speed and body attitude information provided by the GPS / IMU integrated navigation unit.

[0039] Next, the adaptive vehicle state digital twin building module receives the massive amount of sensor data mentioned above. This module's function is not simply data aggregation, but rather using this input to perform real-time correction and parameter updates on a built-in high-fidelity vehicle dynamics model. Through this process, the module can accurately estimate online key time-varying parameters that cannot be directly measured or that change drastically with operating conditions, such as the vehicle's actual total mass, center of gravity height and longitudinal position, and the crucial peak coefficient of adhesion between the tires and the road surface. The output of this step is a highly synchronized and parameter-accurate "digital twin" model of the physical vehicle state.

[0040] Building upon this, the stability risk potential field prediction module receives the refined digital twin model from the module. The core task of this module is to perform forward-looking risk assessment. It utilizes this model, combined with the driver's current intentions (such as steering wheel angle and throttle opening), to conduct a rapid forward simulation over a short prediction timeframe (e.g., 1-2 seconds). The purpose of this simulation is not to compare with a desired value, but to construct a multi-dimensional stability risk potential field that describes the likelihood of future instability. This potential field quantifies different types of instability risks that the vehicle may experience in the future, such as yaw, rollover, and articulated folding, as "potential energy" in a multi-dimensional space. Higher potential energy indicates a greater risk of instability in that region.

[0041] Subsequently, the risk pattern dynamic identification module performs in-depth analysis of the risk potential field generated by the module. This module acts like an intelligent diagnostic expert, focusing not only on the absolute value of the risk potential energy but also on analyzing the topological structure of the potential field in the multidimensional state space, particularly its gradient distribution. By calculating and analyzing the direction and magnitude of the risk potential field gradient, the module can accurately identify the most pressing and dominant instability risk pattern. For example, it can clearly distinguish whether the vehicle is about to fishtail, has a higher tendency to roll over, or whether the articulated parts are showing a tendency to fold uncontrollably.

[0042] Ultimately, the multi-actuator collaborative control module makes adaptive decisions and executes based on the precise risk patterns identified by the module. This is the fundamental difference between this system and traditional control. Instead of using fixed control logic, this module prescribes different "remedies" for different "ailments." It dynamically generates the current optimal control objective and re-plans the priority and coordination methods of all available actuators on the vehicle (including drive motors, service brakes, hydraulic / electric eddy current retarders, etc.). For example, to address yaw instability, it may prioritize driving the torque difference between the wheels; while to address rollover risk, it may prioritize a smooth overall vehicle deceleration. After determining the optimal coordination strategy, the module calculates specific control commands and sends them to the corresponding actuators via the vehicle bus network, thereby eliminating potential instability risks in the most efficient, safest, and least disruptive way.

[0043] Through the repeated application of the above steps, this system can continuously and proactively predict, accurately diagnose, and adaptively coordinate the stability of the vehicle, thereby providing excellent active safety assurance in complex and ever-changing driving environments.

[0044] In a preferred embodiment of the present invention, the specific implementation of the adaptive vehicle state digital twin construction module is described in detail. The core task of this module is to construct an online dynamics model that can synchronize with the physical vehicle in real time and is self-correcting. This model provides an accurate and reliable foundation for all subsequent prediction and control processes. The realization of this task relies on the organic combination of a structured vehicle dynamics model and an efficient online parameter identification algorithm.

[0045] First, a vehicle dynamics model that can fully describe the dynamic behavior of commercial vehicles needs to be established. In this embodiment, the model is constructed as a multi-degree-of-freedom nonlinear state-space equation system, the general form of which is as follows:

[0046] Equations of state:

[0047] Observation equation: y = h(x, u, p)

[0048] The specific meanings of each vector are as follows.

[0049] The specific meanings of each vector are as follows:

[0050] x is the system's state vector, which contains variables describing the vehicle's core motion state, such as the sideslip angle β and the yaw rate ω. r Side roll angle φ, longitudinal speed v x And the articulation angle θ for articulated trains h wait.

[0051] u is the system's control input vector, representing the operation applied by the driver or the system, such as the steering wheel angle δ, and the driving or braking torque applied to the wheels.

[0052] y is the system's observation vector, corresponding to physical quantities that the vehicle sensor network can directly measure, such as the rotational speed of each wheel, longitudinal and lateral acceleration, etc.

[0053] p represents the key parameter vector of the model, which is the core object of online identification in this module. This vector concentrates parameters that have a decisive influence on vehicle dynamics but are easily variable with operating conditions, preferably including the vehicle mass m and the center of gravity height h. g , the distances a and b from the center of mass to the front and rear axles, and the yaw moment of inertia I of the entire vehicle. z And the tire-to-road peak adhesion coefficient μ, which characterizes road surface conditions. peak .

[0054] To address the issue of the constantly changing key parameter vector p during actual operation, this embodiment employs an online estimation algorithm for real-time model correction. Taking the Extended Kalman Filter (EKF) algorithm as an example, its core idea is to augment the parameter vector p to be identified into the original state vector x, forming an augmented state vector X. aug :

[0055] X aug =[x T ,p T ] T

[0056] The augmented state vector X aug The estimation process is accomplished through an iterative prediction-update loop. This algorithm utilizes actual information measured by the sensors to continuously refine the estimation of the augmented state vector, thereby achieving collaborative identification of the state and parameters.

[0057] The prediction steps of the EKF algorithm are as follows:

[0058]

[0059] In this step:

[0060] This represents the prior augmented state estimate predicted at time k based on information from time k-1.

[0061] This represents the optimal estimate of the posterior augmented state after the update is completed at time k-1.

[0062] f(...) is a discretized nonlinear state transition function established based on the vehicle dynamics model.

[0063] Pk∣k-1 Let be the covariance matrix of the prior estimation error.

[0064] A k Let f(...) be the state transition function for the augmented state vector x aug exist The Jacobian matrix is ​​obtained at the given location.

[0065] Q k The covariance matrix of the process noise represents the uncertainty of the model itself.

[0066] The update steps of the EKF algorithm are as follows:

[0067]

[0068] P k|k =(IK k H k )P k|k-1

[0069] In this step:

[0070] z k Let k be the actual measurement vector of the sensor network at time k.

[0071] h(...) is a nonlinear observation function established based on the vehicle model.

[0072] These are theoretical observations obtained based on prior state estimation.

[0073] K k The Kalman gain is used to balance the weights between model predictions and actual measurements.

[0074] The optimal posterior augmented state estimate obtained at time k after incorporating actual measurement information is the final output of this step.

[0075] H k For the observation function h(...) on the augmented state vector X aug exist The Jacobian matrix is ​​obtained at the given location.

[0076] R k The covariance matrix of the measurement noise characterizes the uncertainty of the sensor measurement.

[0077] P k|k This is the updated posterior estimation error covariance matrix.

[0078] Through the continuous iteration of the above prediction and update steps, the adaptive vehicle state digital twin construction module ensures that its internal parameter vector p always closely approximates the actual physical parameters of the vehicle. Finally, this module outputs a high-fidelity digital twin model, calibrated in real time and accurately reflecting the current dynamic characteristics of the vehicle, to the stability risk potential field prediction module. This provides a solid and reliable data foundation for subsequent predictions and decisions by the entire control system.

[0079] In embodiments of the present invention, the stability risk potential field prediction module receives a high-fidelity digital twin model, calibrated in real time, provided by the adaptive vehicle state digital twin construction module. The core function of this module is to shift the perspective of the control system from error correction at the current moment to proactive prediction of potential future risks, thereby laying the foundation for implementing preventative control strategies.

[0080] The module's operation begins with a state trajectory prediction process based on forward simulation. Specifically, the module utilizes the state equations of a refined digital twin model. The system inputs the current vehicle state vector x(t), key parameter vector p(t), and control input vector u(t) determined by the driver's actions. This is then used to predict the future time domain T. p By performing fast numerical integration on the state equation, the system can efficiently solve for a series of trajectory points x(t+τ) representing the future motion state of the vehicle, where θ<τ≤T. p This series of trajectory points constitutes a deterministic prediction of the vehicle's future dynamic behavior.

[0081] Furthermore, the originality of this invention lies in the fact that it does not directly use the predicted trajectory, but instead constructs a multidimensional stability risk potential field U based on the trajectory, capable of quantifying various instability risks. risk The risk potential field is a scalar field defined in the vehicle's state space. Its physical meaning is that the higher the potential energy at a state point, the closer the vehicle is to the instability boundary, and the greater the risk. In a preferred embodiment of the invention, the comprehensive risk potential field is composed of multiple sub-potential fields corresponding to different instability modes, formed by a weighted summation:

[0082] U risk =w yaw U yaw +w roll U roll +w jack U jack

[0083] in:

[0084] U risk This represents the value of the overall risk potential field.

[0085] U yaw U roll U jack These are the yaw risk sub-potential fields, the rollover risk sub-potential fields, and the articulated folding risk sub-potential fields, respectively. yaw w roll w jack These are the weighting coefficients for each sub-potential field. These coefficients can be preset and calibrated according to vehicle type, loading conditions, etc., to adjust the sensitivity to different risk types.

[0086] The specific construction methods for each sub-potential field are as follows:

[0087] Lateral risk sub-potential field U yaw It is used to quantify the risk of a vehicle skidding or fishtailing. It is defined as a normalized quadratic function of the vehicle's yaw rate and sideslip angle relative to its dynamic stability boundary:

[0088]

[0089] Where, ω tr Let β be the yaw rate of the tractor. t The sideslip angle of the tractor unit's center of gravity.

[0090] ω tr,lim and β t,lim These are the dynamic stability limits for yaw rate and sideslip angle, respectively. These two limits are not fixed but depend on the real-time vehicle speed v. x and the tire-road peak adhesion coefficient μ estimated by the module peak The results are obtained through dynamic calculations, thus demonstrating the adaptability to current driving conditions.

[0091] Side rollover risk sub-potential field U roll This is used to quantify a vehicle's rollover tendency. It is based on a key indicator that directly reflects rollover margin—the Lateral Load Transfer Rate (LTR). The LTR is calculated as follows:

[0092] LTR = (F z,right -F z,left ) / (F z,right +F z,left )

[0093] Where F z,right and F z,left These represent the vertical loads on the left and right wheels of the vehicle's coaxial side, respectively. The subpotential field is constructed as an exponential function so that the risk energy increases sharply near the critical state:

[0094] U roll =exp(c roll ·max(0,|LTR|-LTRcrit ))-1

[0095] Among them, LTR crit c is the preset rollover risk warning threshold. roll This is a positive growth coefficient used to regulate the intensity of the potential field growth. This function ensures that the risk potential energy is zero when |LTR| is below the threshold, but grows exponentially once it exceeds the threshold.

[0096] For articulated trains, the articulated folding risk sub-potential field U jack It is used to quantify the risk of runaway folding between the tractor and semi-trailer. It is defined as a normalized quadratic function of the articulation angle and its rate of change:

[0097]

[0098] Where, θ h It is the hinge angle. θ represents the hinge angular velocity. h,lim and These are the safety thresholds for the articulation angle and its angular velocity, set based on the vehicle's geometric parameters and driving speed.

[0099] By applying the above formula to each state point on the predicted trajectory, the stability risk potential field prediction module can ultimately generate a quantitative, multi-dimensional risk potential field distribution map describing the future risk evolution trend. This "future risk map," containing rich risk information, will be fully transmitted to the risk pattern dynamic identification module for subsequent in-depth diagnostic analysis.

[0100] In one specific embodiment of the present invention, the risk pattern dynamic identification module receives a multidimensional stable risk potential field U generated by the stable risk potential field prediction module, which describes the future risk evolution trend. risk The core innovation of this module lies in the fact that it not only assesses the magnitude of risk, but also deeply analyzes the underlying causes of risk through a sophisticated diagnostic method, thereby providing clear decision-making guidance for subsequent adaptive control.

[0101] The implementation of this diagnostic method does not rely on simple threshold comparison, but rather on analyzing the risk potential field U. risk This is accomplished through the topological structure within the vehicle's multidimensional state space. Technically, this manifests as the calculation and analysis of the gradient of the risk potential field. The gradient vector is a mathematically precise tool that describes the direction and rate of change of a scalar field; therefore, it can reveal which one or more state variables' anomalies contribute most to the current sharp increase in risk.

[0102] Specifically, this module first calculates the comprehensive risk potential field U. riskThe gradient vector is obtained by taking the partial derivatives of the key state variables that constitute the state space.

[0103]

[0104] in: Let s be the gradient vector of the risk potential field. i This represents the i-th key state variable constituting the state space. For example, s1 could be the yaw rate w of the tractor. tr s2 can be the lateral load transfer rate (LTR), and s3 can be the hinge angle θ. h And so on. Each component in the gradient vector Its absolute value directly reflects the impact of comprehensive risk on the state variable s. i Sensitivity to change.

[0105] To normalize and compare risk contributions across different dimensions, this embodiment further constructs a normalized gradient energy vector G. norm Each component G of this vector {norm,i} This represents the proportion of the i-th state variable's contribution to the total risk gradient energy. Its calculation method is as follows:

[0106]

[0107] By calculating G norm The vector system can quantitatively determine the main sources of risk growth.

[0108] Based on this, the risk pattern dynamic identification module classifies and identifies the dominant instability risk pattern M according to a preset discrimination logic. This discrimination logic is directly related to G. norm The components with the highest energy in the vector are associated:

[0109] If the state variables related to yaw dynamics (such as the yaw rate w of the tractor) tr And the centroid side slip angle β t The sum of the gradient energy components corresponding to ) in G norm If the dominant risk pattern is in the middle, the system will identify the dominant risk pattern as the yaw instability pattern (M). yaw ).

[0110] If the gradient energy component corresponding to the state variable (such as the lateral load transfer rate LTR) related to vehicle rollover dynamics is in G norm If the dominant risk pattern is in the middle, the system will identify the dominant risk pattern as the rollover risk pattern (M). rollover ).

[0111] If the state variables related to the folding dynamics of articulated trains (such as the articulation angle θ) hand its angular velocity The sum of the gradient energy components corresponding to ) in G norm If the dominant risk pattern is in the middle, the system will identify the dominant risk pattern as the articulated folding pattern (M). jacjjnife ).

[0112] Ultimately, this module will identify the unique dominant risk pattern (e.g., M). yaw M rollover Or M jacjjnife This diagnosis, as a clear and categorized diagnostic conclusion, is passed to the multi-actuator collaborative control module. This diagnosis serves as the direct basis for subsequent adaptive control strategy selection, ensuring that the control system can take the most appropriate intervention measures targeting the root cause of the problem.

[0113] In a preferred embodiment of the present invention, the multi-actuator collaborative control module serves as the final decision-making and execution center of the entire control system. It receives a clear dominant instability risk mode M from the risk mode dynamic identification module. The fundamental innovation of this module lies in the adaptability of its control logic, that is, it can dynamically match the optimal control objective and actuator collaborative strategy according to the "diagnosis" of the risk, thereby achieving precise intervention in a "targeted" manner.

[0114] The operation of this module is primarily manifested as an adaptive matching mechanism for the control strategy. The core of this mechanism lies in the fact that the system's control objectives and the calling logic and priorities of various actuators on the vehicle (e.g., drive system, service braking system, auxiliary braking systems such as hydraulic / electric eddy current retarders) are not static, but are determined in real time by the input risk mode M.

[0115] When the dominant instability risk pattern identified by the risk pattern dynamic identification module is the yaw instability pattern (M) yaw When this happens, this module determines that the most urgent task is to generate a direct yaw correction torque ΔM. z This is to suppress the vehicle's tendency to sideslip or fishtail. The target yaw moment is generated by a proportional-derivative (PD) controller:

[0116]

[0117] Where, ω {tr,d} ω is the desired yaw rate calculated based on the steering wheel angle and vehicle speed. tr k is the actual yaw rate of the vehicle. p and K dThese are proportional and derivative gains, respectively. To achieve this goal, the module will employ a hierarchical actuator invocation strategy to minimize the impact on vehicle ride comfort and energy consumption while ensuring safety: First, the drive system is prioritized to generate yaw moment imperceptibly by applying a small torque difference to the left and right drive wheels (for electric drive axles); Second, if the risk escalates, auxiliary braking devices are smoothly intervened to reduce the overall vehicle speed; Third, only in critical situations will the conventional differential braking (ESC function) be activated to apply braking force to individual wheels to generate the strongest corrective torque.

[0118] When the dominant instability risk pattern is identified as a rollover risk pattern (M) rollover At this point, the control logic of this module will undergo a fundamental change. The primary control objective of the system will no longer be yaw control, but rather the immediate reduction of the root cause of the rollover tendency—lateral acceleration a. y The most direct and effective way to achieve this goal is to reduce the overall vehicle speed. Therefore, the cooperative control strategy will prioritize and smoothly activate auxiliary braking or symmetrical service braking of all wheels to generate a target longitudinal deceleration 'a'. {x,target} The magnitude of this deceleration is positively correlated with the severity of the rollover risk.

[0119] a x,target =f decel (|LTR|-LTR crit )

[0120] Where f decel This is a monotonically increasing function. Meanwhile, to avoid inappropriate unilateral braking force exacerbating body roll, this module will actively and strictly limit or prohibit the intervention of differential braking.

[0121] When the dominant instability risk mode is identified as the articulated folding mode (M jacjjnife This typically occurs on articulated trains. This module will implement a non-obvious, cooperative control strategy designed to actively "stretch" the vehicle assembly. Its control objective is to generate a suppression articulation angle θ. h Divergent equivalent damping torque. To achieve this, the strategy prioritizes braking the semi-trailer's wheels to create a stable deceleration effect at the rear of the vehicle; simultaneously, it maintains or even moderately increases the tractor's driving force. This synergistic "pull-back-pull-forward" action generates an effective "straightening" torque at the articulation point, thereby suppressing further folding tendencies.

[0122] After determining the control objective and actuator invocation strategy, this module needs to define the macroscopic control requirements (such as the target yaw moment ΔM). z and target longitudinal deceleration a {x,target}This is precisely allocated to each actuator at the underlying level. This process is achieved by solving a constrained optimization problem with the following objective function J:

[0123]

[0124] And satisfy physical constraints:

[0125] u act,min ≤u act ≤u act,max

[0126] Where: u act This is the actuator command vector, which contains specific commands such as braking pressure for each wheel, motor torque, and retarder gear position. des For example, [a] is a generalized control objective vector generated based on the current risk model. {x,target} ;ΔM z B is the control efficiency matrix, which describes the contribution of each actuator instruction to the generalized control objective. W v and W u These are the control accuracy weight matrix and the control cost weight matrix, used to balance the "control effect" and the "control cost" (such as energy consumption, wear and tear, and comfort).

[0127] The adaptability of this control allocation method is ultimately reflected in the weight matrix W. v and W u The weight matrices are not fixed values, but are modified in real time based on the identified risk pattern M. For example, in dealing with M... rollover In this mode, the system will significantly increase W. u The element value corresponding to the differential braking command is selected, thus greatly "penalizing" this term during optimization, making it difficult to activate. In this way, the multi-actuator cooperative control module ensures that each intervention is a well-thought-out optimal solution for the current specific risk mode.

[0128] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A longitudinal and lateral coordinated anti-yaw stability control system for commercial vehicles, characterized in that, include: An adaptive vehicle state digital twin building block is used to estimate key time-varying parameters characterizing vehicle dynamics in real time. The stability risk potential field prediction module is connected to the adaptive vehicle state digital twin construction module to predict the multidimensional stability risk potential field of the vehicle in the future prediction time domain based on key time-varying parameters. The risk pattern dynamic identification module, connected to the stability risk potential field prediction module, is used to identify the dominant instability risk pattern by analyzing the topological structure of the multidimensional stability risk potential field. The multi-actuator collaborative control module, connected to the risk mode dynamic identification module, is used to adaptively generate and execute collaborative control strategies based on the dominant instability risk mode.

2. The system according to claim 1, characterized in that, The adaptive vehicle state digital twin construction module is specifically used to: based on the deviation between on-board sensor data and the output of a preset vehicle dynamics model, update the key time-varying parameters in real time through an online estimation algorithm. The key time-varying parameters include at least one of the following: vehicle mass, center of gravity position, and tire-road peak adhesion coefficient.

3. The system according to claim 1, characterized in that, The multidimensional stability risk potential field is composed of a weighted sum of multiple sub-potential fields associated with different instability modes, and its calculation method satisfies the following relationship: U risk =w yaw U yaw +w roll U roll +w jack U jack ; Among them, U risk To comprehensively assess the risk potential, U yaw U roll U jack The sub-potential fields w represent the risks of yaw, tipping, and articulation folding, respectively. yaw w roll w jack These are the corresponding weighting coefficients.

4. The system according to claim 1, characterized in that, The risk mode dynamic identification module is specifically used to: calculate the gradient of the multidimensional stability risk potential field in the vehicle state space, and identify the dominant instability risk mode based on the direction of the main component of the gradient.

5. The system according to claim 4, characterized in that, The dominant instability risk modes include at least one of the following: yaw instability mode, rollover risk mode, and articulated folding mode.

6. The system according to claim 1, characterized in that, The multi-actuator collaborative control module adaptively generates and executes collaborative control strategies by dynamically adjusting the control target and the calling priority of the actuators for different dominant instability risk modes identified.

7. The system according to claim 6, characterized in that, When the dominant instability risk mode is the yaw instability mode, the cooperative control strategy prioritizes calling the vehicle's drive system or differential braking system to generate a direct yaw correction torque.

8. The system according to claim 6, characterized in that, When the dominant instability risk mode is the rollover risk mode, the cooperative control strategy prioritizes the use of the vehicle's auxiliary braking or service braking system to actively reduce the vehicle speed and limits the intervention of differential braking that generates asymmetric braking force.

9. The system according to claim 6, characterized in that, When the dominant instability risk mode is the articulated folding mode, the cooperative control strategy prioritizes braking the semi-trailer wheels while simultaneously maintaining or increasing the tractor's driving force to create a suppression of the articulation angle θ. h A divergent stretching effect.

10. A method for longitudinal and lateral coordinated anti-yaw stability control of commercial vehicles, characterized in that, Includes the following steps: Construct an adaptive digital twin of vehicle state to estimate key time-varying parameters characterizing vehicle dynamics in real time; Based on the key time-varying parameters, the multidimensional stability risk potential field of the vehicle in the future prediction time domain is predicted. By analyzing the topological structure of the multidimensional stability risk potential field, the dominant instability risk mode is identified. Based on the dominant instability risk pattern, a collaborative control strategy is adaptively generated and executed.

Images